2.1. Introduction
Inland valleys constitute over 38% of the total wetlands in the sub-Saharan region and are cropped extensively with lowland rice in the wet season (WARDA, 2008).
The rainfed lowland rice cropping has attributes of non-irrigated with bund fields occasionally flooded during a certain period of time. Low and unstable yields were recorded on about two-thirds of total rainfed lowland rice area due to water shortage during the growing period, flooding and nutrient limitation (Tsubo et al., 2006, Haefele et al., 2006, Samson et al., 2004, Fukai, 1999, Fujisaka, 1990). Yields are strongly influenced by seasonal characteristics as well as by spatial heterogeneity over soil types, topographic sequences and agrohydrologic conditions (Wade et al., 1998).
The topography is the main driver of leaching and soil erosion on one hand and on the other hand it influences the duration of submergence period, resulting in heterogeneity in inherent soil fertility. The soils in areas of higher altitude become less fertile as a result of depletion of nutrients due to runoff which generates in contrast a higher organic carbon and clay content in the soils in the lower position (Homma et al., 2003, Tsubo et al., 2005). These records from rainfed lowland from Asia may be different for West Africa where the use of water control means such as simple bund and short canals constructed by cultivators is less common. Raes et al.
(2007) by a mean of modelling demonstrated that bund could appreciably increase the production of rain-fed lowland rice in Tanzania more in wet year than the normal year. The bund are reported to have benefit to the production by increasing the ponded water depth, regulating the hydric regime and producing increases in grain yield through enhancing fertilizer use efficiency (Touré et al., 2009, Srivastava et al., 2009).
Iron in the soils is also recognized to be another source of variation in rainfed lowland environment. Chérif et al. (2009) confirmed that the iron toxicity is one of the constraints of the cultivated lowland in West African savanna. It occurs on average in more than 50 % of the lowlands and approximately 60 % of cultivated rice fields are affected by this constraint. Fe toxicity produces nutritional disorders associated with a reduction process of Fe3+ into Fe2+ in the flooded conditions. Indeed, nutrient and water management are reported in Becker and Asch (2005) as methods to alleviate the risk of iron toxicity.
Therefore, a good understanding of the yield determining factors in lowlands is a prerequisite for the management in terms of fertilization and water retention. Beside, there is absence of long-term trial on rice crop yields inland valley of West Africa that combines effect of bund and fertilizer. This study examined variation in the
0
production of dry matter and grain yield under lowland conditions in four consecutive years at Dogué inland valley, Benin. The objective of this chapter is to quantify the effect of slope position, bund and fertilizer application on rice yield and to point out the some constraints to rice yield in relation to slope position.
2. 2. Materiel and methods
2.2.1. Site description
The experiment was conducted in a researcher managed on-farm trial located in Dogué village (9°05´N, 01°55´E). The area is located in southern Donga district, North West of Benin Republic (West Africa). The rainfall is presented as mono-modal distribution across the 4 years. Daily weather data were collected from the research climate station of the IMPETUS project at about 1 km from the field. The rainfall pattern is shown in Fig. 2.1. During the growing period from July to November, the rainfall recorded in 2007, 2008, 2009 and 2010 was 793, 833, 690 and 1191 mm, respectively. The onset of the dry season was earlier in 2009 than in the other years.
Figure 2.1: Monthly rainfall in 2007, 2008, 2009, 2010 during the growing period in Dogué village.
2.2.2. Experiment
A spilt plot design was laid out with the combination of three factors: (1) slope position: upslope (up) and downslope (down), (2) fertilizer inputs: with and without mineral fertilizer at a rate of 60kgN and 40kgPha-1 and (3) runoff control (bund): with
Figure 2.2: Experimental layout and treatments in the Dogué field trial in 2007, 2008, 2009 and 2010. Bund and without bund plots are located in the same slope. UpUn: upslope without bund, UpBu: upslope with bund, DoUn: downslope without bund, DoBu: downslope with bund;
+F: with fertilizer, -F: without fertilizer application.
The fertilizer treatment was laid out at random into four replications at combination of bund and slope position each (Fig. 2.2). Subplot size was 5 m x 5 m. Experiments were repeated for four years (2007 to 2010) at the same position for all the plots.
The site is characterized by ferruginous tropical soils in the well drained areas. The slope of 3 % situated between an upland with sandy loams overlying ironstone and the bottom with more hydromorphic and loamy soils. According to FAO soil classification the soils at the upper slope are Lixisols and at the lower slopes Gleysols.
2.2.3. Field management
Chemical and physical soil characteristics were summarized in Table 2.1. Every cropping cycle was separated by a fallow period during the dry season. After clearing and completely removing the fallow vegetation that is grown in dry season, the land was hand ploughed then sown with the lowland rice variety NERICAL-26. The rice
Upslope Downslope Upslope
-F and + F plots
-F and + F
plots -F and + F
plots
-F and + F plots
no bund bund
UpUn UpBu
DoBu DoUn
was dibble seeded at 20 cm x 20 cm spacing and thinning at to 2 plants per hills. The sowing date varied between years: 18, 1, 7 and 3 July in 2007, 2008, 2009 and 2010.
Weeding was carried out when necessary. Harvest was made on 17 Nov., 7 Nov., 6 Nov. and 19 Nov. in 2007, 2008, 2009 and 2010 respectively. All crop residues were removed from the plots after harvest.
Table 2.1: Soil physical and chemical properties of the 0-20 cm layer in Dogué experimental field trial. n is the number of samples. SD is the standard deviation.
Soils properties Unit Upslope Downslope
Mean (n=16)
SD (n=16)
Mean (n=16)
SD (n=16) Physical properties
Fine earth
(elements < 2mm)
% 96.00 4.00 90.00 7.00
Sand % 39.42 - 25.15 -
Clay % 4.10 - 18.50 -
Chemical properties
pH (H2O) - 5.36 0.27 5.63 0.34
Corg % 0.65 0.07 0.93 0.27
Total N % 0.039 0.005 0.064 0.015
Bray P ppm 1.21 0.64 1.76 0.98
CEC cmol kg-1 4.17 0.56 5.53 1.36
K+ cmol kg-1 1.64 0.53 2.36 1.41
Ca2+ cmol kg-1 0.19 0.17 0.23 0.13
Mg2+ cmol kg-1 0.56 0.08 0.69 0.18
Na+ cmol kg-1 0.00 0.00 0.03 0.04
2.2.4. Field measurements and lab analysis
Total aboveground biomass was collected at 38 DAS from two subplots of 1m x 1m.
Leaf samples were extracted for analyses of Fe and N concentration with one repetition per treatment for Fe and two for N in 2007, with two repetitions for both Fe
At maturity, rice grain and total aboveground biomass were obtained. For both plant biomass and grain the sampling area was made of two randomly selected 1m x 1m area. The weight of samples was corrected to the number of hills and the moisture content after 72h oven drying.
Soil samples for each plot (total of 32 plots) were collected in 2006 during the fallow period at up- and down-slope positions from 0 to 20 cm depth. Soil texture was determined using pipette method. Organic carbon estimation was made using Walkley and Black method (1934). The total N in the soil was measured with the Kjeldahl method. The exchangeable bases were extracted with the acetate of ammonium and measured by spectro-photometry with atomic absorption. The Cation Exchange Capacity (CEC) is determined by an extraction with chloride of potassium followed by micro distillation and titrimetry. The assimilable phosphorus was determined by modified method Bray.
During the appearance of ponding water, water level was recorded with a ruler periodically (1 to 3 times in the week) during the cropping season.
2.2.5. Statistical Analysis
Data were analyzed using SAS (Version 9.0). PROC mixed procedure using the Restricted Maximum Likehood method was performed for ANOVA. The model was firstly run with slope position, bund, fertilizer and year factors as main effects.
Random effect concerned the nested effect of bund in position level. Furthermore, the model was run by classifying year. The Tukey test was used and allowed mean separation when the analysis of variance showed a significant factorial effect. We used Pearson correlation coefficients (R) to examine the relationship among grain yield, ponded water level, Fe concentration and N concentration in rice (SAS Institute, 2003). The significance level was fixed at p < 0.05.
2.3. Results
2.3.1. Growth and Grain Yield
Examination of the factors position, bund, fertilizer and year on grain yield, N and Fe in leaves content is made in Table 2.2 for the combined 4 years. The effect of year variation was significant for the three explained variables (grain yield, N leave content and Fe concentration). In addition, bund and fertilizer had significant effect on rice yield. Year to year variation interacted also with the position and fertilizer effects on
grain yield. Fig. 2.3 shows in combination of 4 years, that the highest grain yield was observed in the upper slope position and significantly with bund condition and for fertilizer application. N in plant was only significantly responsive to position level and bund. Position had also significant effect on Fe concentration in addition to many other interactions. The interactions concerned mainly the position with bund, fertilizer and year. The three levels interactions were related to year, position and fertilizer.
Table 2.2: Effects of position (P), bund (B), fertilizer (F) and year variation (Y) on grain yield, N leaf content (N plant) and Fe concentration for 4 years combined. d.f.: degree of freedom;
DDF: denominator degree of freedom of covariance parameters.
ns, not significant at the <0.05 probability level, nd = not determined
Factors d.f DDF F ratio
Grain yield
N plant Fe
concentration
Y 3 84 0.03 <0.0001 <0.0001
P 1 12 ns 0.03 <0.0001
B 1 12 0.03 0.002 ns
F 1 84 0.0001 ns ns
PxB 1 12 ns ns 0.01
BxF 1 84 ns ns ns
FxP 1 84 ns ns <0.0001
BxPxF 1 84 ns ns ns
YxP 3 84 <0.0001 ns <0.0001
YxB 3 84 ns ns 0.04
YxF 3 84 0.03 ns ns
PxBxY 3 84 ns ns ns
FxBxY 3 84 ns ns ns
PxFxY 3 84 ns ns 0.02
FxBxPxY 3 84 ns ns ns
0 1 2 3 4 5 6 7
Up Down Bund No bund Fert No fert
Factor
Grain yield (Mgha-1)
Figure 2.3: Overall trends of factors impact of rice grain yield. Year 2007, 2008, 2009, 2010 are combined. Up and Down refer to upslope and downslope position respectively. Fert and no fert refer to fertilizer and no fertilizer application respectively.
Table 2.3 presents the effect of the three experimental factors on the grain yield, N in plant and Fe concentration at maturity for each year. Grain, N in plant and Fe concentration had diverse responses on bund and slope position during the 4 years of observation. For grain yield, the slope position had a significant effect 2 out of 4 years (2008 and 2010). The bund effect was also significant only in 2007. Fertilizer impact on grain yield started with the two last years. Significance of interaction between factor sources was limited to the position and bund in 2008 and 2010.
In the case of N in plant, there was in addition to position effect in 2008, bund and fertilizer effects in 2008 and 2009, the interaction between position and fertilizer application in 2007. Fe concentration was affected by position in all year except 2009 however in this year, position rather interacted with bund.
Table 2.3: ANOVA table grain yield, N leaf content (N plant) and Fe leaf concentration as function of slope position (P), bund (B) and fertilizer (F) input in 2007, 2008, 2009 and 2010.
Source of variation
Year Grain yield
N plant Fe
concentration
P 2007 ns ns nd
2008 0.001 0.005 0.01
2009 ns ns ns
2010 0.008 ns 0.03
B 2007 0.02 ns nd
2008 ns 0.004 ns
2009 ns 0.005 ns
2010 ns ns ns
F 2007 ns ns nd
2008 ns 0.01 ns
2009 0.0006 0.002 ns
2010 0.02 ns ns
F x P 2007 ns 0.001 nd
2008 ns ns ns
2009 ns ns ns
2010 ns ns ns
P x B 2007 ns ns nd
2008 0.03 ns ns
2009 ns ns 0.0009
2010 0.009 ns ns
B x F 2007 ns ns nd
2008 ns ns ns
2009 ns ns ns
2010 ns ns ns
P x B x F 2007 ns ns nd
2008 ns ns ns
2009 ns ns ns
2010 ns ns ns
ns, not significant at the <0.05 probability level, nd = not determined
0
Up Down Bund No bund Fertilizer No fertilizer Uslope bund Upslope no bund Downslope bund Downslope no bund
Tre atm ent
Table 2.4: Mean grain yield, mean N content and Fe concentration by year in Dogué field trials.
2.3.2. Spatio-temporal evolution of rice production and relationship with N, Fe and water level according to fertilizer bund and position factors
The year 2007 showed the lowest yield during the 4 years of observation (Table 2.4).
Bund operation was the significant factor on yield in this year (Table 2.3). No effect of fertilizer application was recorded but bund contributed to the increase of grain yield (Fig. 2.4). The upslope plots with bund showed slightly higher N concentrations than the downslope plots. However, N was lower in the fertilizer plots in upslope and higher in the fertilizer plots in downslope. In controversy, higher iron content above 800 ppm was recorded in the downslope plots at 38 DAS.
Figure 2.4: Grain yield average under different management practices over 4 seasons.
More grain yield on average was gained in 2008 (Table 2.4). The overall mean N content in plant was increased compared to 2007 (Fig. 2.5). These changes may be responsible for the average grain yield increase in 2008. The factor significance was limited to slope position and its interaction with bund. The plots with bund in
downslope had higher grain yield but no bund plots were higher in upslope (Fig. 2.4).
At 38 DAS, the N content in the plants was higher in the upslope position, with bund and all fertilizer plots (Fig. 2.5). The highest iron concentration at 38 DAS was observed for downslope plots with fertilizer. The value exceeded the threshold of 500 ppm whereas the upslope plots had lower concentrations. The years 2009 and 2010 showed the highest yield (Table 2.4). Fertilizer represented the highest importance in terms of significance level in 2009 for grain yield, N in plant and N uptake (Table 2.5).
The highest N content and N uptake corresponded to the highest yield obtained and correlated as well with the fertilizer application what justifies the level of significance observed with the factor fertilizer in this year. Fe concentration was recorded as the lowest value and is only affected by interaction between position and bund. Position and fertilizer had a significant effect on rice productivity in 2010 and the effect of position was inversed the trends of yielding: the mean grain yield was estimated at 5.2 Mgha-1 in the downslope position, whereas at the upper slope it was 3.8 Mgha-1 (Fig. 2.4).
The impact of bund was observed through accumulation of ponding water during the cropping period (Table 2.6). In all the situations, downslope plots held more water than plots at the upper slope position. The mean ponded water depth was more enhanced by the bund in downslope than in the upslope plots. The water level in upslope plots with bund was particularly high in the year 2010 while highest amount of rainfall was observed. The effect of bund on ponding water started earlier within the first month after sowing. All treatments were significantly different from each other in downslope.
Fertilizer and position interact yearly highly with reference with F ratio in the total experiment (Table 2.2). The effect of fertilizer on Fe concentration in rice is shown per year in Fig. 2.6. For downslope plots there was a trade-off between the fertilizer application and the Fe concentration in 2007 and 2008. However in the upslope plots the Fe risk was associated to the no fertilizer plots in 2007 and 2010.
Relatinship between iron concentration and fertilizer
0 200 400 600 800 1000 1200
2007 2008 Year 2009 2010
Fe Concentration (ppm)
UpO UpF DoO DoF
(a)
(b)
Figure 2.5 : Seasonal evolution of N (a) and Fe (b) proportion according to the different management options. UpBuO = Upslope with bund, no fertilizer, UpBuF =Upslope with bund and fertilizer, UpUnO= Upslope no bund no fertilizer, UpUnF=Upslope no bund with fertilizer, DoBuO= Downslope bund no fertilizer, DoBuF =Downslope with bund and fertilizer, DoUnO=Downslope no bund no bund no fertilizer. DoUnF = Downslope no bund with fertilizer. Values with the same letter within the same year are not significantly different (p=0.05).
Figure 2.6: Effect of fertilizer on Fe concentration at 38 DAS according to the year and the land position. DoO: Downslope without fertilizer, DOF: Downslope with fertilizer, UpO=
Upslope without fertilizer UpF = Upslope with fertilizer.
Table 2.5: Effect of slope, bund and fertilizer on Fe concentration, N in plant content at 38 DAS and N uptake according to the year in Dogué experimental field trial.
Up= Upslope, Do=Downslope, Bu=bund, Un= No bund, Fert= fertilizer, No Fert = no fertilizer
Year Variables 2007 2008 2009 2010
Position Up Do Up Do Up Do Up Do
Fe concentration (ppm)
428 911 320 501 217 195 551 744
N content (%) 1.7 1.6 2.3 1.9 2.5 2.3 2.2 2.0
N-Uptake (kg ha-1) 41 28 85 65 138 112. 62 36
Bund Bu Un Bu Un Bu Un Bu Un
Fe concentration (ppm)
614 725 390 432 196 215 716 579
N plant (%) 1.9 1.4 2.3 1.9 2.6 2.2 2.1 2.1
N-Uptake (kg ha-1) 40 29 80 70 138 112 39 59
Fertilizer Fert No fert Fert No fert Fert No fert Fert No fert
Fe concentration (ppm)
655 684 460 361 197 214 604 691
N plant (%) 1.7 1.6 2.2 2.0 2.6 2.2 2.1 2.2
N-Uptake (kg ha-1) 40 29 94 56 169 81 64 34
Table 2.6: Combined effect of slope and with bund on mean ponded watertable level during the growing period and during 30 DAS (first month of growing cycle), Fe and N concentration in leaves and grain yield. Fe con. refers to leaves Fe concentration at 38 DAS, n is the
The rainfall conditions during the experimental seasons were on average uniform during the first three years but in 2010, total rainfall was above the average. Mean grain yield of the 4 years ranged from 3.81 Mgha-1 to 4.36 Mgha-1.
2.4.1. Effect of land position
Soil characteristics of the experimental field were representative for topography induced soils. The gap in grain yield between the up and downslope was reduced in 2009 and reinversed in 2010 (Fig. 2.4). The higher ponding water depth in early season and across the season in 2010 supported the hypothesis of intensified N-leaching and hence N-losses in 2010 in upslope plots. The land position is associated with fertility: decline of soil fertility is mainly caused by erosion due to the
frequent depletion of N from the upper slope during the rain events. It has been shown differences in soil texture and organic C between upper and lower slope (Table 2.1). This reinforced the hypothesis of erosion occurrence in upper slope because organic C, N and available P are associated with the selective transport of fine aggregates which are chemically richer than the coarser ones (Wan and El-Swaify, 1997). Moreover, the cropping frequency at upslope explains also the loss of organic C and N through an enhanced mineralization and crop export due to historically more frequent cropping activities (Wezel et al., 2002).
2.4.2. Effect of fertilizer application
On average over all treatments, fertilizer application (60kgN and 40kgPha-1) increased yield whereas this increase was not significantly different for the first two years (Table 2.3). The impact of fertilizer has been high in the year 2009 leading to the increase of grain yield by 0.45 Mgha-1 with fertilizer. Boling et al. (2010) found N deficiency in no fertilized plots was responsible for 35%-63% of yield gaps on farmer’s fields in Java. In year 2009, where the strongest effect was recorded and in 2010, the fertilizer resulted in a higher yield at upper slope than in the lower position.
2.4.3. Effect of bund
Bund appears to have in overall experiment duration a positive impact on grain yield although in yearly variation it was only significant in 2007 and interacted with position in 2008 and 2010. The bund was important in maintaining flooded conditions on the plots by preventing runoff and N loss through runoff. The use of water control technology was described by former works to reduce spatial variability in soil water content and to be effective for weeds management (Hayashi et al., 2009). In downslope position, maximum water accumulation seems not to be related to the total rainfall since maximum of ponded water level was obtained in year 2009, recorded as the driest year. The observed fluctuations came in line with the findings of Touré et al. (2009) where the mean ponded water depth in plots with bund increased from valley fringe (0-9 cm) toward valley bottom (2-20cm). In this study,
upslope condition significantly in 2008 and 2009 and at downslope in 2009 (Table 2.6). The same impact of bund was recorded previously by Touré et al. (2009). They described the enhancement of soil temperature that might be higher in upland
upslope condition significantly in 2008 and 2009 and at downslope in 2009 (Table 2.6). The same impact of bund was recorded previously by Touré et al. (2009). They described the enhancement of soil temperature that might be higher in upland